I'm testing with pooler, but I'm get an error: "Connection terminated unexpectedly" on stress test.
I'm testing with pooler, but I'm get an error: "Connection terminated unexpectedly" on stress test. My next test will be without.... thanks!
Dtbh I've found Neon very unreliable in the past with connections.
LYes, I'm discover stranger things on stress scenarios.
AI, personally, think without the pooler will be a better experience
D@AJR As a sanity check, when I use the patch API, what happens to existing connections?
AIn most scenarios, a new pool will be created when you send traffic after the PATCH change to your configuration, and the existing connections will idle and then time out.
Dah dang, so that would deal with if the database password was expiring (since it would trigger a disconnect on the old pools), but wouldn't be useful for @AJR Hmm, I _think_ I _may_ have found a sadly
LHi, I'm working with Drizzle (node-postgres) + Hyperdrive + Neon, I'm trying with/without pooler's connection string and with/without SSL required in connection string... but, sometimes I receive an error: Connection terminated unexpectedly ... any ideas?
LI try with node-postgress pool configuration like: min = 0, idleTimeout=0, etc...
The error appears on stress scenarios (500 request in 10 seconds)
The error appears on stress scenarios (500 request in 10 seconds)
Dbeen thinking about this more; is there really no way to force close existing connections from the Hyperdrive side of things?
AWell, we'd need to build it. It'd be pretty doable, it's just not something we've had to do before.
ABut no, we don't currently have a "nuke this pool" button available to us.
AUnrelated: we've started teeing up migration of existing Hyperdrives to the new pool placement system. I love my demo page screencaps, so here y'all go.
DCould we have have the same/similar page that shows the stats like that for our Hyperdrives?
T@dave | ai.moda what do you think of hyperdrive's current analytics on latency? ie what are you looking to get out of this
Tpr incoming
DI'd be curious what my direct to database origin latency would be for this for example.
AIf you're in an experimenting mood, you can do vanilla TCP connections to your DB from Workers. It usually goes about as well as the demo page suggests so we don't suggest it, but it'd give you the easy comparison.
https://developers.cloudflare.com/workers/tutorials/postgres/
https://developers.cloudflare.com/workers/tutorials/postgres/
Cloudflare Docs
In this tutorial, you will learn how to create a Cloudflare Workers application and connect it to a PostgreSQL database using TCP Sockets and Hyperdrive. The Workers application you create in this tutorial will interact with a product database inside of PostgreSQL.
DThis is a prod Worker, I don't wanna attempt fate. 
Doh I would also like to see number of connections from Hyperdrive to the origin database.
It'd be fun to see "connections" from Workers to Hyperdrive too, but maybe not that useful.
It'd be fun to see "connections" from Workers to Hyperdrive too, but maybe not that useful.
T@Emilio this is what we were discussing
DI can see both parts being interesting especially when combined with transactions, so it's easy to spot if there are bottlenecks there
FHey folks. anyone has had success connecting to a fly.io hosted postgress via a tunnel? 
I setup an app with
But I'm not able to connect hyperdrive to it via the tunnel.
(Following this https://developers.cloudflare.com/hyperdrive/configuration/connect-to-private-database/#3-create-an-access-application-to-secure-the-tunnel)
However...if I setup a public IP and hit that directly... I can?
I setup an app with
cloudflaredBut I'm not able to connect hyperdrive to it via the tunnel.
(Following this https://developers.cloudflare.com/hyperdrive/configuration/connect-to-private-database/#3-create-an-access-application-to-secure-the-tunnel)
However...if I setup a public IP and hit that directly... I can?
Cloudflare Docs
Hyperdrive can securely connect to your private databases using Cloudflare Tunnel and Cloudflare Access.
AInteresting. I haven't tested with fly personally. Are you able to get it working with just psql across an arbitrary TCP tunnel to your DB, no hyperdrive in the loop?
My guess would be that the connectivity from your tunnel egress to your origin database isn't there, so cloudflared can't reach postgres
My guess would be that the connectivity from your tunnel egress to your origin database isn't there, so cloudflared can't reach postgres
Fseems to be the case @AJR
was right now pinging the fly.io folks also https://community.fly.io/t/cloudflared-tunnel-and-postgres/23927/4
if I setup a public IP on the DB, and connect via a xxxx.fly.dev url, i am able to connect with hyperdrive (no tunnel, same credentials).
and SSHing into another different machine in the same internal network, and runnning a psql into the db with the internal URL worksl (
But the issue comes from setting up the tunnel and not being able to hit the DB.
was right now pinging the fly.io folks also https://community.fly.io/t/cloudflared-tunnel-and-postgres/23927/4
if I setup a public IP on the DB, and connect via a xxxx.fly.dev url, i am able to connect with hyperdrive (no tunnel, same credentials).
and SSHing into another different machine in the same internal network, and runnning a psql into the db with the internal URL worksl (
psql postgres://user:pass@database.internal/db-nameBut the issue comes from setting up the tunnel and not being able to hit the DB.
Fly.io
It might help if you could say more about your Postgres instance, why TLS is a hard requirement in this context, and the full details of the connection strings have been tried so far: both the cases that succeeded and the cases that failed. (* out any passwords, of course—and feel free to do the same for any names that you consider sensitive—b...
DDoes using
now()Yes. Any volatile function will disable caching for that query.
AYou can check in the
pg_procprovolatile
IWe have started seeing an increased number of
Error: write CONNECTION_CLOSED <id>.hyperdrive.local:5432
RHi All, any idea when we can expect MySQL support
AASoon!
Dis sthe new pool placement system live yet?
AFor all new or updated Hyperdrives, yes. Also, we've begun backfilling to existing Hyperdrives. If you want yours bumped to the front of the line, just toss the ID here or in a DM and I'm happy to do so.
DIs Hyperdrive a transaction, session, or statement pooler?
ATransaction-mode pooler.
DThank you!
NDoes Hyperdrive maintains connection on BOM or DEL datacenters ?
Our application is facing 3x to 5x higher latency compared to direct connection to the database.
Database is hosted on ap-south-1 (Nearby to BOM datacenter)
For context we have both cloudflare pro and workers paid account.
Our application is facing 3x to 5x higher latency compared to direct connection to the database.
Database is hosted on ap-south-1 (Nearby to BOM datacenter)
For context we have both cloudflare pro and workers paid account.
ASend me the ID for your hyperdrive, let me take a look.
AAh, just dug into this a bit. With legacy regional pooling we were not able to maintain connections in BOM datacenters, but I believe we should be able to lift that restriction for the new global pools. Let me talk to the team about getting that changed in our next release.
In the meantime, if you send me your IDs I will manually place them into our BOM datacenters for you.
In the meantime, if you send me your IDs I will manually place them into our BOM datacenters for you.
NSorry it took time to reply, I really appreciate you helping me with this. Thanks!
I have two hyperdrive config. One got created few months back
Another hyperdrive I have created just now,
Let me know which hyperdrive you are able to place in on
I have two hyperdrive config. One got created few months back
75199f96334c42559f89fe73965875e8Another hyperdrive I have created just now,
7ae2159c8f4e4c77bf037815e02e8995Let me know which hyperdrive you are able to place in on
ANo worries, thank you for bringing this to our attention.
Both of those will now run correctly out of our BOM data centers. Please let me know how it looks for you.
Both of those will now run correctly out of our BOM data centers. Please let me know how it looks for you.
NIt looks amazing,
Latency for the initial request went down from 600-800ms (old hyperdrive config) to 30-50ms.
Latency for the initial request went down from 600-800ms (old hyperdrive config) to 30-50ms.
AYou love to see it.
Good luck building!
Good luck building!
Ahello gys, i want to know if i can use Hyperdrive without workers, with normal nodejs app?
ANo, that is not something we currently support, I'm afraid.
CC @thomasgauvin
CC @thomasgauvin
JI'd like to use Neon with Hyperdrive, but I see conflicting documentation whether to turn on/off Neon's own pooling.
Cloudflare's docs says to turn it off:
https://developers.cloudflare.com/hyperdrive/examples/neon/#:~:text=uncheck%20the%20connection%20pooling
Neon's docs says to turn it on:
https://neon.tech/blog/hyperdrive-neon-faq#so-should-i-use-hyperdrive-together-with-neons-pooling
WHO DO I TRUST
Cloudflare's docs says to turn it off:
https://developers.cloudflare.com/hyperdrive/examples/neon/#:~:text=uncheck%20the%20connection%20pooling
Neon's docs says to turn it on:
https://neon.tech/blog/hyperdrive-neon-faq#so-should-i-use-hyperdrive-together-with-neons-pooling
WHO DO I TRUST
ASince you're asking in the Hyperdrive discord, I vote you trust us
ASerious answer: if it works for you with both, it's fine to have it on. We discourage it because it isn't going to make a ton of difference for most users, and adds a lot of complexity to your traffic flow. In particular the poolers for some hosting services really struggle with prepared statements, and some popular JS/Rust postgres drivers make very heavy use of those.
JThank you for the answer!!
